Binary Computation Offloading in Edge Computing Using Deep Reinforcement Learning

As data-driven applications become increasingly prevalent, traditional cloud computing faces challenges such as latency and operational costs. Edge computing solves these issues by using nearby servers for real-time processing. However, determining the optimal offloading strategy remains complex. This paper investigates a Deep Reinforcement Learning (DRL)-based binary offloading strategy for edge computing in mobile environments. DRL combines reinforcement learning and deep neural networks to adapt to real-time data and diverse environmental conditions. Experimental study demonstrates the effectiveness of the proposed approach over local and remote execution in terms of total overhead and energy consumption.